Esa Prakasa | Indonesian Institute of Sciences (original) (raw)

Papers by Esa Prakasa

This paper presents a visual inspection method based on image processing techniques. The method a... more This paper presents a visual inspection method based on image processing techniques. The method aims to detect surface defects found on pixel chip pads. Pixel chip is a tiny sensor used in Inner Tracking System (ITS) — a large particle detector of ALICE experiment (A Large Ion Collider Experiment). The chips record particle trajectories after collision event in the ITS system. In a mass production stage, the chip quality needs to be accurately inspected and assessed to ensure its technical specification satisfies. The chips are manufactured to build up the ITS system. Considering this large scale production, an inspection system based on imaging techniques is applied to provide fast and accurate chip surface assessment. This paper proposes a method to assess the quality of surface pad by using image processing techniques. The method consists of three main steps. Firstly, K- Means clustering is used to segment the surface pad into clean and defect areas. In the second step, the defect areas are extracted by applying Gabor filter. The last step is conducted by applying Canny Edge filter to detect surface defects on chip pad area. Experimental results show that the proposed method can significantly achieve high accuracy of 84.9% and recall 77.9%.

arXiv (Cornell University), May 27, 2021

A novel approach for designing the next generation of vertex detectors foresees to employ wafer-s... more A novel approach for designing the next generation of vertex detectors foresees to employ wafer-scale sensors that can be bent to truly cylindrical geometries after thinning them to thicknesses of 20-40 µm. To solidify this concept, the feasibility of operating bent MAPS was demonstrated using 1.5 cm × 3 cm ALPIDE chips. Already with their thickness of 50 µm, they can be successfully bent to radii of about 2 cm without any signs of mechanical or electrical damage. During a subsequent characterisation using a 5.4 GeV electron beam, it was further confirmed that they preserve their full electrical functionality as well as particle detection performance. In this article, the bending procedure and the setup used for characterisation are detailed. Furthermore, the analysis of the beam test, including the measurement of the detection efficiency as a function of beam position and local inclination angle, is discussed. The results show that the sensors maintain their excellent performance after bending to radii of 2 cm, with detection efficiencies above 99.9 % at typical operating conditions, paving the way towards a new class of detectors with unprecedented low material budget and ideal geometrical properties.

2018 International Conference on Computer, Control, Informatics and its Applications (IC3INA), 2018

Vegetables are good sources to meet the needs for protein, vitamins, minerals for human. One of p... more Vegetables are good sources to meet the needs for protein, vitamins, minerals for human. One of popular choices for vegetables in Indonesia is Pak Choy (Bassica rapa). Good quality vegetables is usually identified by the color and the shape of the leaves. Therefore an automatic system to detect the quality of the leaves is needed. In this paper, we propose a proper method to detect the quality of the Pak Choy leaves using machine learning. Monitoring the quality of Pak Choy leaves with the naked eye is usually conducted based on the color of the leaves. The healthy leaves are usually characterized by green color while the unhealthy leaves are usually have a green color with the yellow spot. Based on these observations, we develop the system using color intensity features such as RGB and HSI and Support Vector Machine (SVM) as the classifiers. Our system achieves accuracy of 92.5% using linear kernels.

2020 International Conference on Innovation and Intelligence for Informatics, Computing and Technologies (3ICT)

Broiler chicken meat is one of the most widely consumed meat types in Indonesia, this high level ... more Broiler chicken meat is one of the most widely consumed meat types in Indonesia, this high level of consumption makes a lot of consumer demand in the market. However, there was a found seller who sells broiler chicken meat that are rotten. In this study, we develop chicken meat freshness identification using a convolutional neural network algorithm. This study used the image dataset of broiler chicken breasts. There are two categories of chicken meat used in the study, namely, fresh and rotten. The meat images were acquired by using a smartphone camera. For the process of cropping chicken meat images, we use thresholding with the Otsu method and conversion of RGB images to binary images to select the area of RGB images before cropping the images. The chicken meat images were cropped into three sizes and then used as a dataset in the study. The chicken meat image dataset was trained using a simple architecture that was self-made called Ayam6Net, we also used the AlexNet, VGGNet, and GoogLeNet architectures as a comparison. Ayam6Net has the highest accuracy of 92.9%. From the experiment results, we can conclude that using Ayam6Net architecture with dataset 400times400400\times 400400times400 pixels has a better accuracy result compared with other architectures and other sizes image datasets.

2022 1st International Conference on Smart Technology, Applied Informatics, and Engineering (APICS)

IOP Conference Series: Earth and Environmental Science, 2021

Plankton are free-floating organisms that live, grow, and move along with the ocean currents. Thi... more Plankton are free-floating organisms that live, grow, and move along with the ocean currents. This free-floating organism plays important roles as primary producers, they serve as a link to energy transfer, and a factor that regulates the biogeochemical cycles. Indonesia, with almost 60% of its territory covered by the ocean, harbours a wide variety of planktonic species. However, one of the issues within usual planktonic studies is the lack of a fast and accurate method for identifying and classifying the plankton type. Thus, the computer vision methods on microscopic images were proposed to deal with the problem. The classification follows two main steps, detecting plankton location and followed by plankton differentiation. The segmentation algorithm is required to limit the determination area. The present study describes the segmentation methods on fifteen plankton types. The U-Net based architecture was implemented to segment the plankton texture from other objects. The segmenta...

Proceedings of the 2022 International Conference on Computer, Control, Informatics and Its Applications

International Journal of Electrical and Computer Engineering (IJECE)

Under extreme conditions postmortem, dental radiography examinations can play an essential role i... more Under extreme conditions postmortem, dental radiography examinations can play an essential role in individual identification. In forensic odontology, individual identification traditionally compares antemortem dental records radiographs with those obtained on postmortem examination. As such, these traditional methods are vulnerable to oversights or mistakes in the individual identification of unidentified bodies. Digital technology can develop forensic odontology well. An automatic individual identification system is needed to support the forensic odontology process more easily and quickly because there are still opportunities to be created. We aimed to review the complete range of recent developments in identifying individuals from panoramic radiographs. We study methods in gender identification, age estimation, radiographic segmentation, performance analysis, and promising future directions.

2022 16th International Conference on Telecommunication Systems, Services, and Applications (TSSA)

Proceedings of the 2022 International Conference on Computer, Control, Informatics and Its Applications

Proceedings of the 2022 International Conference on Computer, Control, Informatics and Its Applications

Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 2022

A novel approach for designing the next generation of vertex detectors foresees to employ wafer-s... more A novel approach for designing the next generation of vertex detectors foresees to employ wafer-scale sensors that can be bent to truly cylindrical geometries after thinning them to thicknesses of 20-40 µm. To solidify this concept, the feasibility of operating bent MAPS was demonstrated using 1.5 cm × 3 cm ALPIDE chips. Already with their thickness of 50 µm, they can be successfully bent to radii of about 2 cm without any signs of mechanical or electrical damage. During a subsequent characterisation using a 5.4 GeV electron beam, it was further confirmed that they preserve their full electrical functionality as well as particle detection performance. In this article, the bending procedure and the setup used for characterisation are detailed. Furthermore, the analysis of the beam test, including the measurement of the detection efficiency as a function of beam position and local inclination angle, is discussed. The results show that the sensors maintain their excellent performance after bending to radii of 2 cm, with detection efficiencies above 99.9 % at typical operating conditions, paving the way towards a new class of detectors with unprecedented low material budget and ideal geometrical properties.

2020 International Conference on Radar, Antenna, Microwave, Electronics, and Telecommunications (ICRAMET)

The image captured by the surveillance camera at a large distance yields a low-resolution image. ... more The image captured by the surveillance camera at a large distance yields a low-resolution image. Commonly, researchers recognize a face from a distance by improving the quality of the low-resolution image. After the image has been upgraded, it will be compared with the high-resolution face image saved on the database. But in this paper, the inverse is applied. The model is made by downgrading the quality of the high-resolution image in order to make it similar to the low- resolution image. Some methods in digital image formation are used to make the model. Some experiments also conducted to compare the model with images obtained by the real cameras at various distances. In order to optimize the model, some parameters were used to tune some steps of low-resolution modeling, i.e., scaling, kernel size of the filter, gamma, and compression quality. The result shows that the proposed model can improve the recognition performance on SC face.

Khazanah Informatika : Jurnal Ilmu Komputer dan Informatika

Traffic accidents are generally caused by human error as a driver. The main cause is that the veh... more Traffic accidents are generally caused by human error as a driver. The main cause is that the vehicle shifts away from the driving lane without the driver realizing it. Usually, because the driver is sleepy or drunk. Therefore, it is necessary to have a system that functions to assist the driver's navigation to stay on the correct driving path, such as a driver assistance system (DAS). In this system, the driving lane detector is the main part. This system serves to assist the driver's navigation to stay on the correct driving path. Vehicles are installed with cameras to record video towards the road ahead. Computers are also installed for image processing, identifying left and right road lines, and forming ego-lane. This paper offers an image processing-based method for recognizing driving lanes and presenting visualizations in real-time. This method has been tested using a data set, that video driving on Indonesian highway on the Cipali and Palikanci sections using dashboa...

Applied Computational Intelligence and Soft Computing

Hybrid maize seed production is a relatively complex task due to the coexistence of three distinc... more Hybrid maize seed production is a relatively complex task due to the coexistence of three distinct types of maize plants in the field: female, male, and contaminant/off-type plants. Female and contaminant/off-type plants’ tassels should be removed immediately following flowering initiation, while male tassels should be retained to allow cross-pollination between male and female plants. Therefore, development of an intelligent tassel classification system is deemed critical for hybrid purity decision-making. The research’s primary contribution is the integration of two widely used transfer learning architectures, Inception V3 and SqueezeNet, with stacking ensemble machine learning using four algorithms (logistic regression, support vector machine, random forest, and k-nearest neighbors) for rapid classification of tassel images. Tenfold cross-validation was used to evaluate the model performance. Cloud computing was also investigated using EfficientNet to compare the predictive perfo...

2018 International Conference on Computer, Control, Informatics and its Applications (IC3INA), 2018

Parking space availability has become common problem in many big cities. This problem occurs due ... more Parking space availability has become common problem in many big cities. This problem occurs due to fast growing of vehicle ownership. Therefore, the demand of parking area in big cities is also increased. An information system on parking space availability may help the driver to find accurately the parking location. This real-time system can avoid the drivers waste their time in looking the available parking space. This paper aims to implement Local Binary Pattern (LBP) as a method for extracting distinguishable features of the vacant and occupied parking slot. Support Vector Machine (SVM) classifier is used to differentiate the status of parking slot, either vacant or occupied. Combination of LBP and SVM has been tested on a total of 7, 670 sample images. Validation result shows that the proposed algorithm can provide a high accuracy, 96.8%, in classification of parking slot availability.

This paper presents a visual inspection method based on image processing techniques. The method a... more This paper presents a visual inspection method based on image processing techniques. The method aims to detect surface defects found on pixel chip pads. Pixel chip is a tiny sensor used in Inner Tracking System (ITS) — a large particle detector of ALICE experiment (A Large Ion Collider Experiment). The chips record particle trajectories after collision event in the ITS system. In a mass production stage, the chip quality needs to be accurately inspected and assessed to ensure its technical specification satisfies. The chips are manufactured to build up the ITS system. Considering this large scale production, an inspection system based on imaging techniques is applied to provide fast and accurate chip surface assessment. This paper proposes a method to assess the quality of surface pad by using image processing techniques. The method consists of three main steps. Firstly, K- Means clustering is used to segment the surface pad into clean and defect areas. In the second step, the defect areas are extracted by applying Gabor filter. The last step is conducted by applying Canny Edge filter to detect surface defects on chip pad area. Experimental results show that the proposed method can significantly achieve high accuracy of 84.9% and recall 77.9%.

arXiv (Cornell University), May 27, 2021

A novel approach for designing the next generation of vertex detectors foresees to employ wafer-s... more A novel approach for designing the next generation of vertex detectors foresees to employ wafer-scale sensors that can be bent to truly cylindrical geometries after thinning them to thicknesses of 20-40 µm. To solidify this concept, the feasibility of operating bent MAPS was demonstrated using 1.5 cm × 3 cm ALPIDE chips. Already with their thickness of 50 µm, they can be successfully bent to radii of about 2 cm without any signs of mechanical or electrical damage. During a subsequent characterisation using a 5.4 GeV electron beam, it was further confirmed that they preserve their full electrical functionality as well as particle detection performance. In this article, the bending procedure and the setup used for characterisation are detailed. Furthermore, the analysis of the beam test, including the measurement of the detection efficiency as a function of beam position and local inclination angle, is discussed. The results show that the sensors maintain their excellent performance after bending to radii of 2 cm, with detection efficiencies above 99.9 % at typical operating conditions, paving the way towards a new class of detectors with unprecedented low material budget and ideal geometrical properties.

2018 International Conference on Computer, Control, Informatics and its Applications (IC3INA), 2018

Vegetables are good sources to meet the needs for protein, vitamins, minerals for human. One of p... more Vegetables are good sources to meet the needs for protein, vitamins, minerals for human. One of popular choices for vegetables in Indonesia is Pak Choy (Bassica rapa). Good quality vegetables is usually identified by the color and the shape of the leaves. Therefore an automatic system to detect the quality of the leaves is needed. In this paper, we propose a proper method to detect the quality of the Pak Choy leaves using machine learning. Monitoring the quality of Pak Choy leaves with the naked eye is usually conducted based on the color of the leaves. The healthy leaves are usually characterized by green color while the unhealthy leaves are usually have a green color with the yellow spot. Based on these observations, we develop the system using color intensity features such as RGB and HSI and Support Vector Machine (SVM) as the classifiers. Our system achieves accuracy of 92.5% using linear kernels.

2020 International Conference on Innovation and Intelligence for Informatics, Computing and Technologies (3ICT)

Broiler chicken meat is one of the most widely consumed meat types in Indonesia, this high level ... more Broiler chicken meat is one of the most widely consumed meat types in Indonesia, this high level of consumption makes a lot of consumer demand in the market. However, there was a found seller who sells broiler chicken meat that are rotten. In this study, we develop chicken meat freshness identification using a convolutional neural network algorithm. This study used the image dataset of broiler chicken breasts. There are two categories of chicken meat used in the study, namely, fresh and rotten. The meat images were acquired by using a smartphone camera. For the process of cropping chicken meat images, we use thresholding with the Otsu method and conversion of RGB images to binary images to select the area of RGB images before cropping the images. The chicken meat images were cropped into three sizes and then used as a dataset in the study. The chicken meat image dataset was trained using a simple architecture that was self-made called Ayam6Net, we also used the AlexNet, VGGNet, and GoogLeNet architectures as a comparison. Ayam6Net has the highest accuracy of 92.9%. From the experiment results, we can conclude that using Ayam6Net architecture with dataset 400times400400\times 400400times400 pixels has a better accuracy result compared with other architectures and other sizes image datasets.

2022 1st International Conference on Smart Technology, Applied Informatics, and Engineering (APICS)

IOP Conference Series: Earth and Environmental Science, 2021

Plankton are free-floating organisms that live, grow, and move along with the ocean currents. Thi... more Plankton are free-floating organisms that live, grow, and move along with the ocean currents. This free-floating organism plays important roles as primary producers, they serve as a link to energy transfer, and a factor that regulates the biogeochemical cycles. Indonesia, with almost 60% of its territory covered by the ocean, harbours a wide variety of planktonic species. However, one of the issues within usual planktonic studies is the lack of a fast and accurate method for identifying and classifying the plankton type. Thus, the computer vision methods on microscopic images were proposed to deal with the problem. The classification follows two main steps, detecting plankton location and followed by plankton differentiation. The segmentation algorithm is required to limit the determination area. The present study describes the segmentation methods on fifteen plankton types. The U-Net based architecture was implemented to segment the plankton texture from other objects. The segmenta...

Proceedings of the 2022 International Conference on Computer, Control, Informatics and Its Applications

International Journal of Electrical and Computer Engineering (IJECE)

Under extreme conditions postmortem, dental radiography examinations can play an essential role i... more Under extreme conditions postmortem, dental radiography examinations can play an essential role in individual identification. In forensic odontology, individual identification traditionally compares antemortem dental records radiographs with those obtained on postmortem examination. As such, these traditional methods are vulnerable to oversights or mistakes in the individual identification of unidentified bodies. Digital technology can develop forensic odontology well. An automatic individual identification system is needed to support the forensic odontology process more easily and quickly because there are still opportunities to be created. We aimed to review the complete range of recent developments in identifying individuals from panoramic radiographs. We study methods in gender identification, age estimation, radiographic segmentation, performance analysis, and promising future directions.

2022 16th International Conference on Telecommunication Systems, Services, and Applications (TSSA)

Proceedings of the 2022 International Conference on Computer, Control, Informatics and Its Applications

Proceedings of the 2022 International Conference on Computer, Control, Informatics and Its Applications

Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 2022

A novel approach for designing the next generation of vertex detectors foresees to employ wafer-s... more A novel approach for designing the next generation of vertex detectors foresees to employ wafer-scale sensors that can be bent to truly cylindrical geometries after thinning them to thicknesses of 20-40 µm. To solidify this concept, the feasibility of operating bent MAPS was demonstrated using 1.5 cm × 3 cm ALPIDE chips. Already with their thickness of 50 µm, they can be successfully bent to radii of about 2 cm without any signs of mechanical or electrical damage. During a subsequent characterisation using a 5.4 GeV electron beam, it was further confirmed that they preserve their full electrical functionality as well as particle detection performance. In this article, the bending procedure and the setup used for characterisation are detailed. Furthermore, the analysis of the beam test, including the measurement of the detection efficiency as a function of beam position and local inclination angle, is discussed. The results show that the sensors maintain their excellent performance after bending to radii of 2 cm, with detection efficiencies above 99.9 % at typical operating conditions, paving the way towards a new class of detectors with unprecedented low material budget and ideal geometrical properties.

2020 International Conference on Radar, Antenna, Microwave, Electronics, and Telecommunications (ICRAMET)

The image captured by the surveillance camera at a large distance yields a low-resolution image. ... more The image captured by the surveillance camera at a large distance yields a low-resolution image. Commonly, researchers recognize a face from a distance by improving the quality of the low-resolution image. After the image has been upgraded, it will be compared with the high-resolution face image saved on the database. But in this paper, the inverse is applied. The model is made by downgrading the quality of the high-resolution image in order to make it similar to the low- resolution image. Some methods in digital image formation are used to make the model. Some experiments also conducted to compare the model with images obtained by the real cameras at various distances. In order to optimize the model, some parameters were used to tune some steps of low-resolution modeling, i.e., scaling, kernel size of the filter, gamma, and compression quality. The result shows that the proposed model can improve the recognition performance on SC face.

Khazanah Informatika : Jurnal Ilmu Komputer dan Informatika

Traffic accidents are generally caused by human error as a driver. The main cause is that the veh... more Traffic accidents are generally caused by human error as a driver. The main cause is that the vehicle shifts away from the driving lane without the driver realizing it. Usually, because the driver is sleepy or drunk. Therefore, it is necessary to have a system that functions to assist the driver's navigation to stay on the correct driving path, such as a driver assistance system (DAS). In this system, the driving lane detector is the main part. This system serves to assist the driver's navigation to stay on the correct driving path. Vehicles are installed with cameras to record video towards the road ahead. Computers are also installed for image processing, identifying left and right road lines, and forming ego-lane. This paper offers an image processing-based method for recognizing driving lanes and presenting visualizations in real-time. This method has been tested using a data set, that video driving on Indonesian highway on the Cipali and Palikanci sections using dashboa...

Applied Computational Intelligence and Soft Computing

Hybrid maize seed production is a relatively complex task due to the coexistence of three distinc... more Hybrid maize seed production is a relatively complex task due to the coexistence of three distinct types of maize plants in the field: female, male, and contaminant/off-type plants. Female and contaminant/off-type plants’ tassels should be removed immediately following flowering initiation, while male tassels should be retained to allow cross-pollination between male and female plants. Therefore, development of an intelligent tassel classification system is deemed critical for hybrid purity decision-making. The research’s primary contribution is the integration of two widely used transfer learning architectures, Inception V3 and SqueezeNet, with stacking ensemble machine learning using four algorithms (logistic regression, support vector machine, random forest, and k-nearest neighbors) for rapid classification of tassel images. Tenfold cross-validation was used to evaluate the model performance. Cloud computing was also investigated using EfficientNet to compare the predictive perfo...

2018 International Conference on Computer, Control, Informatics and its Applications (IC3INA), 2018

Parking space availability has become common problem in many big cities. This problem occurs due ... more Parking space availability has become common problem in many big cities. This problem occurs due to fast growing of vehicle ownership. Therefore, the demand of parking area in big cities is also increased. An information system on parking space availability may help the driver to find accurately the parking location. This real-time system can avoid the drivers waste their time in looking the available parking space. This paper aims to implement Local Binary Pattern (LBP) as a method for extracting distinguishable features of the vacant and occupied parking slot. Support Vector Machine (SVM) classifier is used to differentiate the status of parking slot, either vacant or occupied. Combination of LBP and SVM has been tested on a total of 7, 670 sample images. Validation result shows that the proposed algorithm can provide a high accuracy, 96.8%, in classification of parking slot availability.